Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
A semiempirical model has been developed for the estimation of the normal boiling points of organic
compounds. The normal boiling point is calculated as the ratio of the enthalpy of boiling to the entropy of
boiling. Both these values are estimated independently using a combination of additive group contribution
and nonadditive molecular descriptors. A group contribution model is proposed for the estimation of the
enthalpy of boiling, based on the experimental data for 1322 structurally diverse organic compounds. The
average absolute error associated with this estimation is 0.83 kJ/mol. A semiempirical model has been developed
for the estimation of the entropy of boiling using experimental entropy data for the entire set. This model is
a modification of Trouton’s rule and takes into account the effect of hydrogen bonding and molecular flexibility.
The average absolute error associated with this estimation is 4.7 J/(K mol). The average absolute error associated
with the estimation of the normal boiling point, calculated as a ratio of the enthalpy of boiling to the entropy
of boiling, for the entire data set is 9.3 K (2.3%). A good correlation is observed between the estimated and
the experimental boiling points with an r2 of 0.98. This model is easy to use, accurate, and applicable to a
wide variety of organic compounds.
∆Sb ) ∆SbTrans + ∆SbRot + ∆SbConf (3) Table 1. Summary of the Average Absolute Errors for the
Estimation of ∆Sb
∑nibi
terminal atoms and sp hybrid atoms do not contribute to
∆Hb ) (5) flexibility, they are not counted. The rotation of hydrogen atoms
is also ignored. Aliphatic cyclic groups are counted as single-
where ni is the number of times a group appears in the
ring systems. Compounds with a negative value for τ are
compound and bi is its contribution to the enthalpy of boiling.
assigned a value of zero.
The experimental value of the enthalpy of vaporization at
Hydrogen-Bond Parameter (HBP). The effect of hydrogen
298 K (∆Hv) is more commonly available. Our data indicates
bonding was considered as a function of the number of different
that, although ∆Hv values may be more accurate as the
hydrogen-bonding groups [-OH, -COOH, and -NH (including
measurements are performed at lower temperature, these values
-NH and -NH2)] present on the molecule. The HBP is
are less constant than ∆Hb values. Thus, in this study, a group
calculated using the equation proposed by Sanghvi and
contribution scheme is developed for the estimation of ∆Hb
Yalkowsky:16
instead of ∆Hv. The two values are related by the following
equation,
HBP ) x[(#OH) + (#COOH) + 0.0625(#NH)] (10)
∆Hb ) ∆Hv + ∫298 ∆Cp,b dT
Tb
(6) where # is the number of respective groups (aliphatic or
aromatic).
where the value of ∆Cp,b is the difference between the enthalpy- Estimation of the Enthalpy of Boiling. The compounds were
temperature slopes of the gas phase and the liquid phase. Since fragmented using the group breakdown scheme (Table 2)
2858 Ind. Eng. Chem. Res., Vol. 45, No. 8, 2006
is attached to two or more doubly or triply bonded groups. ar: aromatic; bp: atoms that are part of two aromatic rings; br: atoms that are part of more than
one aromatic rings; ring: atoms that are part of an aliphatic ring system; and ring- 3,4,5,>5: three, four, five, or more than five-membered aliphatic ring
systems, respectively. 1°: Primary aliphatic; 2°: secondary aliphatic; 3°: tertiary aliphatic; IHB-5,6,7: five-, six-, or seven-membered ring systems, respectively,
formed due to the formation of intramolecular hydrogen bonding. G-2,3,4: Carbons bound to two, three, or four halogens, respectively. Italicized values are
based on <5 occurrences.
Ind. Eng. Chem. Res., Vol. 45, No. 8, 2006 2859
proposed by Jain.19 This scheme takes into account the effect Table 3. Summary of the Average Absolute Errors for the
of individual groups and their proximity on the heat of transition. Estimation of ∆Hb
For describing the structure of compounds used in this study, compounds n data range (kJ/mol) AAEa
132 group values were required. This includes nine new groups rigid-non-H-bonding 287 9.7-48.5 1.0
that were defined to describe the structure of some compounds flexible-non-H-bonding 863 21.6-60.3 0.8
that are not covered by the scheme of Jain et al. The coefficients rigid-H-bonding 40 24.2-50.5 0.8
of the groups for the estimation of enthalpy of boiling were flexible-H-bonding 132 26.6-64.0 0.7
total data set 1322 9.7-64.0 0.8
obtained by multiple linear regression (SPSS version 10) of the
experimental ∆Hb values.
a AAE ) ∑ (|(experimental value - predicted value)|)/n.
Data. The experimental ∆Hv values for 1322 compounds
were obtained from Chickos and Acree.18 These compounds
were selected as the enthalpy measurements were performed at
298 K ((5 K). The experimental ∆Hb* values for these
compounds were calculated using eq 7. The experimental Tb
values for the compounds were obtained from the literature.
The experimental ∆Sb values were calculated as a ratio of ∆Hb*
and Tb. These values were used to develop the model for the
estimation of ∆Sb. The apparent experimental ∆Hb values were
obtained by multiplying the estimated ∆Sb and experimental
Tb values. These values were used to develop the model for the
estimation of ∆Hb.
Estimation of the Boiling Point. The boiling points of the
compounds were calculated by dividing the estimated ∆Hb Figure 1. Estimated versus experimental boiling point (K).
values by the respective estimated ∆Sb values.
to reduce the feasibility of the second hydrogen-bond forma-
Results and Discussion tion. Molecules containing only hydrogen-bond acceptor groups
and no donor groups cannot undergo hydrogen-bond forma-
Entropy of Boiling. On the basis of the statistical analysis, tion. Thus, the presence of groups such as carbonyl, nitro, or
the following equation was obtained for the estimation of ∆Sb: halo does not appear to affect the entropy term. A mathemati-
cally significant effect of such groups in the estimation was
∆Sb ) 87 + 0.35(τ) + 15(HBP) (11) not observed.
Enthalpy of Boiling. The group values for the estimation of
The average absolute error of the estimation is 4.7 J/(K mol), the enthalpy of boiling are given in Table 2. The summation of
corresponding to 5.2%. Table 1 summarizes the results obtained these values is used to calculate the total enthalpy for each
using eq 11. The compounds have been divided into four compound using eq 5.
subsets: rigid-non-hydrogen-bonding; flexible-non-hydrogen- The group breakdown scheme of Jain19 was proposed
bonding; rigid-hydrogen-bonding; and flexible-hydrogen- originally for estimating the enthalpy of melting. An addition
bonding. For comparison, the results obtained using Trouton’s of nine new groups was made to describe the structures of some
rule are included in the table. compounds that were not covered by the original scheme. The
The table demonstrates the general applicability of Trouton’s average absolute error associated with the estimation is 0.83
rule, particularly of rigid-non-hydrogen-bonding compounds. kJ/mol, which correspond to 2.3%. These results are summarized
However, the errors associated with flexible compounds are in Table 3. The considerable accuracy obtained with the
slightly lower, and those for hydrogen-bonding compounds are estimation of enthalpy of boiling using this scheme indicates
substantially lower with eq 11. The intercept of 87 J/(K mol) is that the influence of a group on the process of melting and
very close to Trouton’s constant entropy of boiling [88 J/(K boiling is qualitatively similar. However, the coefficients for
mol)], which accounts for the gain in the translational freedom the estimation of enthalpy of boiling are generally larger than
upon boiling. those for the estimation of enthalpy of melting.
The effect of increase in conformational and rotational Normal Boiling Point. The estimated boiling points for the
freedom cannot be totally neglected. These motions are some- compounds used in this study were calculated by incorporating
what restricted in the liquid phase, particularly for flexible eqs 5 and 11 into eq 1:
molecules. Intermolecular hydrogen bonding also restricts
molecular rotations in the liquid state. These restrictions are Tb ) [ ∑nibi]/[87 + 0.35(τ) + 15(HBP)] (12)
relieved upon boiling, since the likelihood of hydrogen bonding
is negligible in the gas phase. Intuitively, the effect of dif- The average absolute error of estimation is 9.3 K, corresponding
ferent hydrogen-bonding groups on the entropy of boiling to 2.3%. A correlation coefficient (r2) of 0.98 was obtained with
depends on their strength. This is reflected in eq 10 which is a standard deviation of 9.8 K. A plot of the predicted versus
used to calculate HBP, where the coefficients of -COOH the experimental boiling points is given in Figure 1.
and -OH groups are larger than that of an -NH group, since It is noticed that slightly higher negative errors were associ-
the former are stronger hydrogen-bonding groups than the ated with high-boiling compounds. The systematic overpredic-
latter. Since the -SH group is a very weak hydrogen-bonding tion for such compounds may be due to the underestimation of
group, its effect on the entropy of boiling is insignificant. The entropy of boiling. Also for extremely high-boiling compounds,
effect of hydrogen bonding increases with the number of such the likelihood for thermal decomposition exists at temperatures
groups, but this increase is not linear. The square root term in near their boiling points.
eq 10 accounts for the fact that the formation of the first The results obtained for the estimation of the boiling points
hydrogen bond imposes geometric restrictions on the molecule are presented in Table 4. The proposed model performs well,
2860 Ind. Eng. Chem. Res., Vol. 45, No. 8, 2006
Conclusion
Absolute error: 5.7 K (1.3%)
The proposed semiempirical model enables a simple, accurate, Supporting Information Available: The experimental and
and widely applicable method for estimation of the normal the estimated values for the normal boiling point along with
boiling point of organic compounds based on their chemical the absolute errors are available for all the organic compounds
structure. The enthalpy of boiling can be estimated using the used in this study. This material is available free of charge via
proposed group contribution values. The entropy of boiling can the Internet at http://pubs.acs.org.
be estimated using a modified Trouton’s rule. The normal
boiling point is than calculated as the ratio of the two values. Literature Cited
(1) Yalkowsky, S. H.; Dannenfelser, R. M.; Myrdal, P.; Simamora, P.
Examples Unified Physical Property Estimation Relationship (UPPER). Chemosphere
1994, 28, 1657-1673.
(a) 1-penten-3-ol. (2) Fisher, C. H. Boiling Point Gives Critical Temperature. Chem. Eng.
1989, 96, 157-158.
(3) Satyanarayana, K.; Kakati, M. C. Note: Correlation of Flash Points.
Fire Mater. 1991, 15, 97-100.
(4) Sladkov, B. Estimation of Molar Volume of Inorganic Liquids at
Boiling Points. J. Appl. Chem. 1991, 64, 2273.
(5) Herberger, K. Discrimination between Linear and Non-Linear Models
Describing Retention Data of Alkylbenzene in Gas-Chromatography.
Chromatographia 1990, 29, 275.
(6) Rucker, G.; Rucker, C. On topological indices, boiling points, and
Experimental Tb: 387.5 K cycloalkanes. J. Chem. Inf. Comput. Sci. 1999, 39, 788-802.
(7) Dearden, J. C. Quantitative Structural-Property Relationship for
Prediction of Boiling Point, Vapor Pressure, and Melting Point. EnViron.
SP3: 2, SP2: 1, ring: 0, #(OH): 1 Toxicol. Chem. 2003, 22, 1696-1709.
∆Sb ) 87 + 0.35(τ) + 15(HBP) (8) Joback, K. G.; Reid, R. C. Estimation of pure-component properties
from group contributions. Chem. Eng. Commun. 1987, 57, 233-243.
(9) Constantinou, L.; Gani, R. New group contribution method for
) 87 + 0.35(2 + 0.5 + 0 - 1) + estimating properties of pure compounds. AIChE J. 1994, 40, 1697-
15(1) ) 102.53 J/(K mol) 1710.
Ind. Eng. Chem. Res., Vol. 45, No. 8, 2006 2861
(10) Stein, S. E.; Brown, R. L. Estimation of Normal Boiling Point from (15) Myrdal, P. B.; Yalkowsky, S. H. Estimating Pure Component Vapor
Group Contribution. J. Chem. Inf. Comput. Sci. 1994, 34, 581-587. Pressures of Complex Organic Molecules. Ind. Eng. Chem. Res. 1997, 36,
(11) Nannoolal, Y.; Rarey, J.; Ramjugernath, D.; Cordes, W. Estimation 2494-2499.
of Pure Component Properties. Part 1. Estimation of Normal Boiling Point (16) Sanghvi, R.; Yalkowsky S. H. Estimation of Heat Capacity of
of Non-Electrolyte Organic Compounds via Group Contribution and Group Boiling for Organic Compounds. Ind. Eng. Chem. Res. 2006, 45, 451-453.
Interactions. Fluid Phase Equilib. 2004, 226, 45-63. (17) Dannenfelser, R. M.; Yalkowsky, S. H. Predicting the Total Entropy
(12) Wessel, M. D.; Jurs, P. C. Prediction of Normal Boiling Points of of Melting: Application to Pharmaceutical and Environmentally Relevant
Hydrocarbons from Molecular Structure. J. Chem. Inf. Comput. Sci. 1995, Compounds. J. Pharm. Sci. 1999, 88, 722-724.
35, 68-76. (18) Chickos, J.; Acree, W. E. Enthalpies of Vaporization of Organic
(13) Basak, S. C.; Gute. B. D.; Grunwald, G. D. A Comparative Study and Organometallic Compounds 1880-2002. J. Phys. Chem. Ref. Data
of Topological and Geometrical Parameters in Estimating Normal Boiling 2003, 32, 519-878.
Point and Octanol/Water Partition Coefficient. J Chem. Inf. Comput. Sci. (19) Jain, A. Ph.D. Thesis, University of Arizona, Tucson, AZ, 2005.
1996, 36, 1054-1060. ReceiVed for reView November 17, 2005
(14) Kolska, Z.; Ruzicka, V.; Gani, R. Estimation of the Enthalpy of ReVised manuscript receiVed February 15, 2006
Vaporization and the Entropy of Vaporization for Pure Organic Compounds Accepted February 16, 2006
at 298.15 K and at Normal Boiling Temperature by a Group Contribution
Method. Ind. Eng. Chem. Res. 2005, 44, 8436-8454. IE051277G